Nothing can be said in this conclusion that is more valuable than the actual data itself. If for some reason you skipped the individual charts, jumping right to the conclusion, you’ll likely not grasp what this story is all about: balance.
For each chart, we recommended a minimum level of CPU and GPU needed to play the game at that particular resolution. By tallying results of all 28 tests (seven games x four resolutions), we’ll summarize how often each solution was able to reach our targeted level of performance.
A word of warning here: while valuable in many ways, this chart has the potential to be misused. Our minimum recommendations are just that--minimums--and only a guideline for offering a blend of image quality and playable performance as affordably as possible.
Often this cheapest acceptable solution was still way out of balance and clearly would benefit from adding more CPU to the mix. The most accurate picture of the level of hardware necessary to truly be balanced can not be portrayed by generalization or summaries, but only by looking at the individual charts for each game and resolution. Also consider that these test systems were clean and only running the essentials. Additional background applications and multi-tasking would give us all the more reason to step up to a higher-end CPU with more muscle.
The dual-core Pentium E6300 managed to deliver playable performance in each game except for Crysis and Grand Theft Auto IV. However, this 2.8 GHz CPU was rarely in balance with the graphics cards, even at playable settings. Adding 200 MHz, a faster FSB, and three times the L2 cache, the Core 2 Duo E8400 faired far better in these games, only failing in Crysis when a 1920x1200 resolution required a match-up with the GeForce GTX 295. That being said, the Radeon HD 4870 X2 and all three Nvidia GeForce cards still often required a quad-core processor, such as the Core 2 Quad Q9550, to be balanced in these games. The Core i7-920 didn’t necessarily beat out the Q9550 in minimum targets reached, but the individual charts still depict many performance advantages garnered by stepping up to this architecture.
The 512MB Radeon HD 4850 offered playable performance all the way up to 1920x1200 in Fallout 3 and Race Driver: GRID, but fell shy in each of the other games, even at the lowest tested resolution. If we later make a switch in these genres to Need for Speed: Shift and Risen, would this once-mighty card then fall completely below “max” settings, even at the lowest resolution? The step up to a GeForce GTX 260 is big, and it brought five of the games to playable levels at 1920x1200. Fallout 3 was even playable at 2560x1600. But the card was reduced to the lowest resolution in S.T.A.L.K.E.R., and still fell below our target completely in Crysis.
The Radeon HD 4890 added playability in Crysis at the lowest resolution, plus 2560x1600 in GTA IV and GRID. However, the most notable improvement for the Radeon HD 4890 was its relatively high level of performance and balance when paired with the dual-core CPUs. When matched up to a quad-core processor, the GeForce GTX 285 was often able to flex its muscles and pass by the Radeon HD 4890, while also adding 1680x1050 S.T.A.L.K.E.R. and 2560x1600 Far Cry 2 playability to the mix.
There is a common misconception that multi-GPU solutions are unable to scale well at low resolutions. This can be cleared up by looking at how the Radeon HD 4870 X2 and GeForce GTX 295 often walk away from the single-GPU cards, even at 1280x1024. While SLI/CrossFire driver support and scaling do vary by individual game, you’ll notice in today’s results that if the single-GPU cards are keeping up, it is typically tied into the sharp-sloped lines depicting CPU limitations.
But let’s be realistic, very few people spend $350 or more on a graphics card to game on a 19” LCD or a lower non-native resolution. Where these cards currently best prove their worth is at higher resolutions. The Radeon HD 4870 X2 handles all but two of the games at 2560x1600, but is reduced to 1920x1200 in S.T.A.L.K.E.R. and 1650x1050 in Crysis. This dual-GPU card offers high performance and is often well-balanced with the E8400 and Q9550, but at times still trailed behind the Radeon HD 4890 when paired with too weak a CPU.
However, this is nothing compared to the GeForce GTX 295, which, when paired with too weak a CPU brings unbalance to a whole new level. The mighty GeForce GTX 295 reaches the highest overall frames per second in the vast majority of the charts, proving it is the most powerful graphics card we tested today. When it is paired with a quad-core CPU, it delivers playable performance in every situation except at 2560x1600 in Crysis and S.T.A.L.K.E.R. But it certainly is not the graphics card to pair with a budget dual-core processor, being outright crippled by the Pentium E6300, it dropped from first place down into last.
Later on, we’ll see how overclocking these platforms changes the story, but first we’ll return in Part 2 of the series to see how these graphics cards perform when paired with dual-, triple-, and quad-core AMD Phenom II processors.
Stay on the Cutting Edge
Join the experts who read Tom's Hardware for the inside track on enthusiast PC tech news — and have for over 25 years. We'll send breaking news and in-depth reviews of CPUs, GPUs, AI, maker hardware and more straight to your inbox.
Wow, this is an amazingly in depth review! I kinda feel that its sponsered by Asus or Corsair, but I guess you kept with the same brand for the sake of controls etc.Reply
I see a great series comingReply
Very helpful stuff.Reply
I'd like to see some discussion on the availability of sub $400 (at times as low as $280) 28" monitors. At this price range, does it make more sense to spend more on the LCD even if less is spent initially on graphics? I would think the benefit of 28" vs. 22" is so great that the extra money could be taken from, say, a 9550 + 4890 combo and getting a 8400/6300 + 4850 instead, with the right motherboard a second 4850 later will pass a 4890 anyway.
I like the balance charts. It's a good way to characterize the data. This article is well constructed and well thought-out.Reply
That being said - is there a way we can compile this data and compute an "optimized" system for the given hardware available? Finding the true, calculated sweet spot for performance/$ would be so nice to have on hand every quarter or twice a year. I'll have to think about this one for a while. There may be some concessions to make, and it might not even work out. But it would be so cool.
Nice one.Now i want to see part two.Reply
I feel like the person that did this review got it finished alittle bit late. I can only assume he did all the testing some months back and has only just finished writing up his results. But its sad to not see the new P55/i5 Systems, AMD Athlon II Quad Cores, or the Radeon 5000 series.Reply
Good review, but hopefully it can be updated soon with some of the newer equipment thats out, to turn it into a fantastic guide for people.
Great job, this is the information people need to be seeing; the way people provide benchmarks these days hardly tells the story to most of the readers. It's definitely important to point out the disparities in ones CPU choice, rather than just assuming everyone uses the i7 all the sites choose. ;)Reply
Looking forward to part 2.
wow! Awesome works! Can't wait til you guys get to the ATI 5000 series. Keep it up! :)Reply
This will take up some of my time. Even though I know how, it's nice to get a refresher every now and then.Reply
you are really kidding me by not considering the ATI 5000 series, although am a fan of nvidia , but this is not fair !Reply